Even had London burned down for good in 1940, as almost happened during the Docklands firestorm of 7 September that did not, in our world, quite conflagrate into chymical marriage like Dresden, 1940 would have already been saved, if by saved we mean known. There may be no other single year in world history—certainly no other single season—more intensely anticipated (see below), more intensely studied in the present tense of unfolding events (see below), and more intensely relived (see below) than the last months of 1940, that great season when Saint Paul's did not burn, not even during the terrible rain of incendiaries at the end of December, the historical moment Connie Willis focuses on in "Fire Watch" (1982), the first of her time travel tales to invigilate this thronged passage in the story of the West. Chastely and reverently, "Fire Watch" treats as unchangeable a series of events already so densely told that to manipulate a jot or tittle of the Memory Cathedral would seem blasphemous, for the heart of the Blitz is precisely the way it happened. There is no theoretical Jonbar Point—unlike, say, the moment when Abraham Lincoln is or is not assassinated—whose enactment would not dishonour what actually came down: when a million life events threatened to petrify in the glare: the lives of all who lived then incinerating into dominos of fire: the medusa clatter of dominos toppling as Hitler Wins, creating a world more terrible even than the stygian nightmare foretold by Katharine Burdekin in Swastika Night (1937): but a transmogrification that by all that is holy did not happen, which is one of the miracles we have lived by in a bad century.

Willis is perfectly capable of doing postcard to the past when she can get away with it—she is not hugely faithful to the abyssal back-lit fragments of historical record that are all we retain of the time of the Black Death, where Doomsday Book (1992), the next instalment in what one might call her Oxford Historians sequence, is set; and when her Oxford gang of history postgrads and docents from the middle of the twenty-first century swans back to the 1880s in To Say Nothing of the Dog (1997), they arrive in an Edenic England that was a nostalgic fiction then, and a lie now—and perhaps before divagating into what's wrong and right about her vast new novel I should jump my gun and say that Blackout/All Clear, one of whose intricately welded, hugely drawn-out climaxes magically transects the events of "Fire Watch," also honours the consensus of the veritable world. It is, in fact, all about honouring that consensus.

Blackout/All Clear (Willis's publisher broke the slow, sometimes infuriatingly seamless crescendo structure of this extremely long novel into two volumes; my proof copy of the second volume boorishly describes it as a "follow-up") is in fact a time travel story not about changing the world but about saving the world; Blackout/All Clear (hence All Clear, because that is what it is all about in the end) is as reverent for this world as Kim Stanley Robinson's Galileo's Dream. But before we glance at All Clear in slightly more detail, I think we need to consider for a moment what may have been Willis's greatest challenge in writing her huge tale, which is the Memory Cathedral Problem: how does an author honour consensus when so much is already known that honour is no longer an emotion appropriate in general to the circumstances, but more centrally an obligation to get it right? Aristotle may have said that fiction is truer than history (anticipating by a mere few thousand years my own excited claim that any Story is inherently non-mimetic), but it does seem that any tale set in a period of history denser than the concinnity magnets of Story can harmonize must be conceived as honouring a reversal of that proposition: for the Blitz is truer than Story.

If this is true of any nonfantastic fiction set then, it is a principle far more poignantly binding on an author of fantastika, where Story is inherently exposed in all its workings (absolute Story being absolute fantastika). A tale like Sarah Waters's The Night Watch (2006), which plays delicately with forms of the supernatural whose incursion is unlikely to change the story of the world, can incorporate a few shortcuts or errors without much risk, but All Clear, being a time travel story driven for almost 1100 pages by the anxiety (see below) that history might be/might have already been changed by the bumbling incursion of the Oxford postgrads, is hugely more vulnerable to scrutiny. And scrutinizing the text for signs of alternative realities is exactly the task any reader familiar with time travel tales, as almost all of Willis's readers will be, will take as having been demanded by the author, on the assumption that what seems to be an "error" in the tale will almost certainly serve as an exact marker that another story of the world is sliding into view.

So, given the almost supernatural density of our knowledge of 1940, it's clear Willis faced an extremely daunting challenge in ensuring that All Clear meant what it said, within reasonable limits. So despite the fact that it is a time travel story, I did not look for errors; though inevitably I found a few, I didn't think the honour of the book—its obedience to the world—depended upon absolute accuracy, nor were its weaknesses exposed by mistakes: rather the reverse in fact. All the same, it's clear that the irradiated density of our knowledge of 1940 must have been a challenge and a burden. There's no point explicating here the complexity of access points we are able to use in 2010 to get a sense of 1940. Among recent studies that do that job for us, the best may be Juliet Gardiner's The Blitz: The British Under Attack (2010) which magisterially exploits various archives, some of them vast; makes narrative sense of this flood of stuff; closes with a ten page bibliography that scratches the surface just deeply enough to evoke chaos theory in the timid. But some of the historical record does need to be mentioned here, as it shines an odd light on the extremely peculiar research missions that Willis's modestly incompetent crew seem to have been assigned. Without getting in harm's way (we will return to that astonishing directive a bit later), their task is to insert themselves into and to report back on ordinary life as it was lived just before and during the Blitz. Whether or not these missions may seem implausibly low-key for a clearly very expensive, cutting-edge project located at a point that SF writers and readers almost certainly (not without cause) think of as Planetary Crisis Time, there is another problem Willis must solve if she is going to honour the time, as I have already said I think she has intended to. The problem, of course, is the problem of the known.

One of the cast, Eileen, becomes a kind of nanny taking care of refugee children in a stately home; another, Polly, becomes a shopgirl. Seriously dislocated by "slippage"—all Willis's time travel stories seem to take off from the fact that time travel does not actually work very well—a third protagonist, whose name is sometimes Mike, sees much of southern England from below. In due course, the three of them do observe quite a bit of the life of the masses; their slow, sidling, sometimes palsied progress through volume one in search of each other, a task bedevilled by slippages and failures of portals back to 2060 to open when they should, gives Willis plenty of opportunities to tell us what they've seen, observing at all times (in order to honour the time) the historical record, which is to say she must have herself confirmed in the historical record anything that might have already been mentioned. And here's the rub. The problem with the first half of All Clear is that the job Willis's cast has been sent back at astronomical cost to accomplish was already done before they got there. What they find out, we have already told them.

In 1937 two men—Charles Madge, who was deeply involved, as was his mentor W. H. Auden, in 1930s Modernism-for-the-people; and Tom Harrisson, an extremely acute and pragmatic sociologist—founded an organization called Mass Observation (it still exists) whose goal was precisely to observe the behaviour of "ordinary" folk in England at a time of growing anxiety about the future; to measure and record the real life of the English people before dread made them sick as war neared, the World War Two everyone in 1937 anticipated with depths of anxiety we hardly feel even now, fingernails clawing at our own edge of things. Mass Observation was founded, in other words, to save 1930s England for the future by learning it. To a surprising degree, the project succeeded. Fifteen books of observations and findings were published by the early 1940s, some of them widely distributed; in 2010 (half a century before our Carry On Boffin historians are sent back to the same time to do the same job) the entire Mass Observation Archive at the University of Sussex near Brighton is available to researchers. True, Mass Observation may reflect a seriously unAmerican sub specie eternitatis Scientific Romance perspective on the events of history; and Connie Willis—though her main accomplishment has been a set of novels that do sound in synopsis as though they were pure Scientific Romance—may not leave the veriest secular/clerical trace of a Wellsian echo in the way she actually tells story, and anyway I've no idea if she actually made use of Mass Observation: but if she did read the books, or access the Archive, every sip of knowledge from this resource must have been attar from a poison chalice: because everything she could have learned would necessarily have added essential detail to the honour she so clearly wished to pay England and the period she so transparently loved: everything Mass Observation had already recorded closed down upon her ability to manoeuvre, to unleash her cast to behave like typical SF protagonists (which they surely do not) and find stuff out. Willis, as she proves in every one of her 1100 pages of honourable crochet-work, knows that they must not fabricate a novum out of the prelapsarian bounty of the real recorded world that we deem holy, and in the event they learn nothing that we don't already know: 1940 is not to be alternated with.

There is a beauty of observance in this, what one does in a church when one obeys the rules. This sense of a gradually unfolding litany did in the end work a sea change on my own initially frustrated reading of the first half of All Clear, the 500 pages of which only slowly confessed themselves, through foggy spirals of story-stuff generated by the cast's unavailing attempts to accomplish anything (see below), as comprising a great extended exercise in obedience not novum. There were points, all the same, when it might have helped to foreground the fact that nothing new could be allowed in church, perhaps through a scene or two in which Oxford historians in 2060 are seen accessing research material already online half a century earlier in the real world. But just as the hordes of Mass Observation observers make no appearance in All Clear neither does—it is one instance out of many—The London County Council Bomb Damage Maps 1939-1945 (London Topographical Society and London Metropolitan Archives, 2005), an edition by Ann Saunders of the 110 Ordinance Survey maps upon which, beginning in September 1940, the London County Council Architect's Department recorded and made available the specific location and degree of bomb damage inflicted throughout the period of the Blitz and later. Willis, who plausibly consulted these hugely detailed maps herself, might have profitably portrayed honcho boffin Dunworthy using them to avoid boners in his attempts to work out exactly where and when bombs fell in 1940, in order to instruct his students where and when never to go: anywhere, that is, they might encounter danger, anywhere the rough beast of history might slouch Godzilla-like across the Channel, anywhere the real world, in its anguish, might need eyes to witness.

Indeed, the least useful pages of All Clear are spent tracing its cast's ultimately baulked attempts not to see anything, and it does take a while to grasp the beauty of All Clear, the intense humility of its portrait of London as her cast increasingly ignores Dunworthy's strictures, especially in two superb, hugely extended setpieces: one devoted to the terrible first bombing raid on 7 September; the second massively expanding on the events first depicted in "Fire Watch" as Saint Paul's almost burns at the end of December. Almost certainly some bad mistakes leak into the text (how else, given the oceans of data she had to attempt to master); but I for one found nothing to complain about. The main errors I noted myself were in fact easily correctible: Willis seems to have consulted a contemporary map of the London Underground, which seems to have led her to assume that the Victoria and the Jubilee Lines, both constructed decades later, were there in 1940; nurses bewilderingly tell patients their temperature in centigrade; and the term "disinformation" seems not to have existed before 1955, the first year it was used to describe false information created, usually by a government, for purposes of deceit. But none of these slips opened any plausible gulf into the alternate realities whose potential irruption haunts her cast. All Clear is a song of London, a song of England, and she has gotten the song right.

That the first volume of the huge tale is almost all recitative with no arias may unfairly deter some readers (at points it came pretty close to deterring me). It took some time to realize that—over and above the exigencies of story—this recitative was actually the song of London clearing its throat. But those exigencies were a burden. Hundreds of pages are spent following the three main protagonists miss every possible connection in their attempts to link up with each other, none of them realizing that these failures are far from coincidental, for the first 700 pages of the story of All Clear are an intercalated rat's nest of Appointments in Samarra tropes where, as it seems, every attempt they make not to interfere with space-time leads them towards places where they commit ever deeper damage. Time and again they run from Death and meet Death smiling thanks. Or so they seem to think. They are wrong.

The actual story of All Clear only begins to make much sense on page 311 of the second volume, when Polly begins to realize what readers from 2010—who have clearly read a lot more time travel stories than Willis's time travelers—will have suspected for some time: that the Oxford Historians have gotten it all back to front. They have feared that their forced presence in 1940, long after they were supposed to find their portals and vamoose back up the time stream, has caused a fatal deterioration in the "entire space-time continuum," whose Gaia-like struggle to maintain the continuity of the world they have fatally compromised; that the Samarra dance they have led is a sort of body English of the death-throes of our world, the world in which Hitler loses the war. This is what Polly thinks she has understood when finally she realizes that

everything had conspired against them, from Theodore's refusal to leave the pantomime to the blocked streets which had kept them from getting here before he left this morning. It was as if the entire space-time continuum had been engaged in an elaborate plot to keep them from reaching John Bartholomew [the protagonist of "Fire Watch"]. Just as it had kept her and Eileen from finding each other last autumn. "How all occasions do inform against us," she thought.

The important thing here is that—after 700 pages of Feydeau-like story teases that seemed to go nowhere, all conducted with the same school-marmish, leadenly-light whimsy that scuppered To Say Nothing of the Dog—we have finally been given a grammar to sort All Clear out. Polly may still have it all back-to-front, but it is now only a matter of time before she and her colleagues use this exposed grammar of story to unpack the truth: that the "entire space-time continuum" has in fact been forcing them to make precisely the interventions in the world of 1940 that they have been trying to avoid; that instead of blasphemously thinning the "weight of the world" until it caves in, their actions have comprised a series of tipping points whose effect is to avoid the more likely prime reality, where Hitler Wins. Once the cast understands this, light pours from the heavens, and the novel ends in elation and Return.

The Appointments in Samarra have been with Life.

Galileo unburnt, London unburned: worlds to pray for.